基于三维降噪自编码器的高光谱变化检测算法  

Hyperspectral change detection algorithm based on three-dimentional denosing autoencoder

作　　者：冯收 陈勇奇 樊元泽 宿南 闫奕名 赵春晖[1,2] FENG Shou;CHEN Yongqi;FAN Yuanze;SU Nan;YAN Yiming;ZHAO Chunhui(College of Information and Communication Engineering,Harbin Engineering University,Harbin 150001,China;Key Laboratory of Advanced Marine Communication and Information Technology,Ministry of Industry and Information Technology,Harbin Engineering University,Harbin 150001,China)

机构地区：[1]哈尔滨工程大学信息与通信工程学院,黑龙江哈尔滨150001 [2]先进船舶通信与信息技术工业和信息化部重点实验室,黑龙江哈尔滨150001

出　　处：《实验技术与管理》2023年第2期7-13,共7页Experimental Technology and Management

基　　金：黑龙江省教育科学“十四五”规划2022年度重点课题(GJB1422105);哈尔滨工程大学教学改革项目(JG2021B0805,JG2021Y064);国家自然科学基金(62002083,61971153,62071136);中央高校基本业务费(3072021CFT0801,3072022QBZ0805,3072022CF0808)。

摘　　要：随着高光谱图像变化检测技术的可实用性越来越高,为提高变化检测算法对高光谱图像中空间信息的关注度,设计了一种基于三维降噪自编码器的高光谱变化检测算法。该算法在特征提取过程中,采用三维卷积层同时提取高光谱图像的空间和光谱信息;此外,还在输入数据中加入高斯噪声并在网络中引入残差网络,用于提高网络的特征提取能力和鲁棒性。实验结果表明,与其他5种对比算法相比,所提出的算法在3种不同数据集上在检测性能上更具优势,不仅能检测出较为连续的变化区域,而且具有一定的泛化能力、更高的检测精度和鲁棒性。With the increasing practicability of hyperspectral image change detection technology,in order to improve the attention of change detection algorithm to spatial information in hyperspectral image,a hyperspectral change detection algorithm based on three-dimensional denosing autoencoder is designed.In the process of feature extraction,three-dimensional convolution layer is used to extract the spatial and spectral information of hyperspectral image at the same time.In addition,Gaussian noise is added to the input data and residual network is introduced into the network to improve the feature extraction ability and robustness of the network.The experimental results show that compared with the other five comparison algorithms,the proposed algorithm has better detection performance on three different datasets.It can not only detect more continuous change regions,but also has certain generalization ability,higher detection accuracy and robustness.

关 键 词：高光谱图像 变化检测 自编码器 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]